Chain-of-Thought Consultation — When to Use CoT Prompting

I will assess your task and deliver a written CoT consultation covering whether chain-of-thought prompting will improve your results, the right CoT variant for your task (zero-shot CoT, few-shot CoT, or self-consistency CoT), how to elicit reasoning in your specific LLM, and whether extended thinking models are a better fit than CoT prompting. Chain-of-thought prompting improves performance on tasks requiring multi-step reasoning — mathematical problems, logical deductions, causal analysis, and complex classification with justification — but has no benefit and increases cost on tasks that don't require intermediate reasoning steps, where direct answer prompting performs equally well at lower token cost.

The consultation covers your task reasoning requirements, CoT variant recommendation, elicitation prompt design for your LLM, self-consistency sampling recommendation if applicable, and extended thinking model comparison.